Mitigating SUV uncertainties using total body PET imaging

Area of Science:

• Nuclear Medicine
• Radiopharmacology
• Medical Imaging Physics

Background:

• Standardized uptake values (SUV) are crucial for quantifying 18F-FDG uptake in PET imaging.
• SUV measurements can be affected by uncertainties and errors, impacting diagnostic accuracy.
• Long-axial field-of-view (LAFOV) PET/CT systems offer potential for image-based quality control.

Purpose of the Study:

• To develop and validate image-based quality control (QC) methods for 18F-FDG PET.
• To derive 18F-FDG activity and patient weight directly from total body (TB) and half-body (HB) PET images.
• To reduce SUV uncertainties and improve the precision of quantitative PET measurements.

Main Methods:

• Regression fits were established using 25 patient scans from a LAFOV PET/CT system to derive image-based activity and weight.
• These regression fits were applied to 56 independent 18F-FDG PET scans (TB and HB) for validation.
• The impact of image-derived values on liver SUVmean and lesion SUVpeak precision was assessed.
• Regression fits were also tested on 20 scans from a short-axial field-of-view (SAFOV) PET/CT system.

Main Results:

• Image-derived 18F-FDG activity and weight strongly correlated with reported values (r > 0.98).
• Accurate derivation of activity and weight was achieved within 4.9% and 3.2% for TB scans and within 4.9% and 3.1% for HB scans, respectively.
• Image-derived values reduced variability in liver and lesion SUV measurements compared to reported values.
• SAFOV system scans yielded comparable accuracy for activity and weight derivation (within 6.7% and 4.5%).

Conclusions:

• 18F-FDG activity and weight can be accurately determined from both TB and HB PET scans using image-derived methods.
• Image-derived values enhance SUV precision and correct for lesion SUV errors, leading to more reliable quantitative measurements.
• Incorporating image-derived values as a QC measure is recommended for reproducible and trustworthy PET uptake quantification.